Bootstraps taxometrics. Solving the classification problem in psychopathology

Area of Science:

• Psychopathology research
• Applied mathematics
• Quantitative psychology

Background:

• Classification in psychopathology is a fundamental empirical question concerning latent structure.
• Current diagnostic systems often rely on conventions or clinical impressions rather than empirical data.
• Distinguishing between taxonic (categorical) and nontaxonic (dimensional) structures is crucial for accurate classification.

Purpose of the Study:

• To present and validate taxometric procedures for determining the latent structure of psychopathology.
• To provide methods for estimating key parameters such as base rates and validity/false-positive rates.
• To advocate for taxometric analysis as a basis for revising diagnostic systems.

Main Methods:

• Utilized two taxometric procedures: MAMBAC and MAXCOV-HITMAX, for independent testing of the taxonic hypothesis.
• Employed 'coherent cut kinetics,' a general approach involving multiple consistency tests for robust inference.
• Required no gold standard criterion, using fallible diagnostic indicators only for initial candidate identification.

Main Results:

• Demonstrated that taxometric procedures can accurately estimate taxon base rates and latent means.
• Showed satisfactory accuracy in estimating valid and false-positive rates for various diagnostic cut-offs.
• Provided confidence in inferring taxonic structure and numerical accuracy through multiple consistency tests.

Conclusions:

• Taxometric analysis offers a rigorous, data-driven method for classifying psychopathology.
• These mathematical procedures provide reliable estimates essential for diagnostic system development.
• Future revisions of diagnostic systems should prioritize taxometric findings over non-taxometric or committee-based decisions.